The purpose of this controlled randomized 10 week trial is to compare the differential effectiveness of phenelzine (monoamine oxidase inhibitor (MAOI)). Cognitive therapy (CT), and pill placebo in reducing symptoms of outpatients with unipolar nonpsychotic major depressive disorder who meet Liebowitz et al.'s (1984, 1988) criteria for atypical depression. This study is important because atypical depression is a common, chronic, disabling illness that is responsive to treatment. Published findings suggest that atypical depression may be especially responsive to phenelzine (approximately 71%). Our pilot data suggest that approximately 85% of atYpical depressions respond to CT. If the response rate of atypical depressions treated with CT equals that of phenelzine, CT would represent an alternative treatment without the significant risks and side effects of an MAOI. This study is the first controlled trial evaluating the effectiveness of a psychotherapy for atypical depression although in practice many of these patients receive psychotherapy. This study is also the first comparison of CT and a MAOI. Minor aims of the study include gathering preliminary data on: the effect of a diagnosed personality disorder on treatment, the characteristic profile of treatment responders, and the durability of treatment effects. One hundred twenty five male and female outpatients. Aged 20-65, with unipolar nonpsychotic atypical depression will be diagnosed and randomized to one treatment condition following a 14-day drug wash-out and exposure to a baseline run-in of a nonspecific treatment. Only patients whose symptoms fail to remit are randomized and treated. Manipulation checks on the adequacy of treatment include platelet inhibition levels and Cognitive Therapy Scale (CTS) scores. Dependent measures cover the following four areas: psychiatric diagnosis, symptom severity cognitions, and interpersonal functioning. The efficacy of treatment is assessed at randomization, weeks 2,4,8 and 10 or at end-point by a clinical evaluator blind to treatment assignment Multivariate, univariate and post hoc analyses are planned.